CS 501: Software EngineeringFall 2000

Lecture 14

System Architecture I

Data Intensive Systems

2

Administration

Midterm examination, October 16, 7:30 to 8:30 p.m.

-- Hollister B14 (note change of room)

-- See course notices for sample questions

3

System Architecture

The overall design of a system:

• Computers and networks (e.g., monolithic, distributed)

• Interfaces and protocols (e.g., http, CORBA)

• Databases (e.g., relational, distributed)

• Security (e.g., smart card authentication, SSL)

• Operations (e.g., backup, archiving, audit trails)

• Software environments (e.g., languages, source control tools)

4

Data Intensive Systems

Examples

• Electricity utility customer billing

• Telephone company call recording and billing

• Car rental reservations (e.g., Hertz)

• Stock market brokerage (e.g., Charles Schwab)

• Web sales (e.g., Amazon.com)

5

Example 1: Electricity Utility Billing

First attempt:

Data input Master fileTransaction Bill

Each transaction handled as it arrives.

6

Criticisms of First Attempt

Where is this first attempt weak?

The requirements have not been specified!!!

7

Transaction Types

• Create account / close account

• Meter reading

• Payment received

• Other credits / debits

• Check cleared / check bounced

• Account query

• Correction of error

• etc., etc., etc.,

8

Typical Requirements

• All payments to be credited on day received

• Customers must be able to query account by telephone

• Cutting off service for non-payment requires management authorization

• Data input staff should process n transactions per day per person

• Error rate must be below 0.01%

• System available 99.9% of business hours

9

Batch Processing: Validation

Data input

Master file

Edit & validation

read only

errors

Validated transactions

Incoming transactions

10

Batch Processing: Master File Update

Master fileupdate

Bills

Validated transactionsin batches

Sort by account

errors Reports

Instructions

11

Benefits of Batch Updating

• All transactions for an account are processed together

• Backup and recovery have fixed checkpoints

• Better management control of operations

• Efficient use of staff and hardware

12

Online Inquiry

Data input Master fileTransactions Bills

read only

Customer service

13

Example 2: A Small-town Stockbroker

• Transactions

Received by mail or over telephone

For immediate or later action

• Complex customer inquiries

• Highly competitive market

14

A Database Architecture

Database(s):

• Customer and account database

• Financial products (e.g., account types, pension plans, savings schemes)

• External databases (e.g., stock markets, mutual funds, insurance companies)

15

Database Architecture

Customer & account database

Products & services database

External services

16

Real-time Transaction

Customer & account database

Products & services database

External services

Real-time transactions

17

Real-time Transactions & Batch Processing

Customer & account database

Products & services database

External services

Real-time transactions

Batch processing

Data input

18

Architectural considerations

• Real-time service during scheduled hours + batch processing overnight

• Combine information from several databases

• Database consistency after any type of failure

two-phase commitreload from checkpoint + logdetailed audit trail

• How will transaction errors be avoided?

• How will transaction errors be corrected?

19

Example: Merger of Two Banks

Each bank has a database with its customer accounts. The databases are used by staff at many branches and for back-office processing.

The requirement is to integrate the two banks so that they appear to the customers to be a single organization and to provide integrated service from all branches.

20

Merger of Two Banks: Options

???

???

AA BB

21

Merger of Two Banks: Architectural Options

I. Convert everything to System A.

convert databasesretrain staffenhance System A (software and hardware)discard System B

II. Build an interface between the databases in System A and System B.

III. Extend client software so that it can interact with either System A or System B database.

22

Distributed Computing: General Problem

An application that is running on one computer wishes to use data or services provided by another:

• Network connectionprivate, public, or virtual private networklocation of firewalls

• Protocolspoint-to-point, multicast, broadcastmessage passing, RPC, distributed objectsstateful or stateless

• Quality of service

23

Network Choices

Public Internet:

Ubiquitous -- worldwideLow cost

Private network:

SecurityPredictable performanceChoice of protocols (e.g., IBM's SNA)

24

Quality of Network Services

Performance

Maximum throughputVariations in throughputReal-time media (e.g., audio)

Business

SuppliersTrouble shooting and maintenanceUpgrades

25

Firewall

Public network

Private network

Firewall

A firewall is a computer at the junction of two network segments that:

• Inspects every packet that attempts to cross the boundary

• Rejects any packet that does not satisfy certain criteria, e.g.,

an incoming request to open a TCP connectionan unknown packet type